CHAPTER 2 SYSTEM PLATFORM
2.2 T WO - RAY C HANNEL M ODEL
In this section, our two-ray channel model of CIR and CFR is shown in Figure 2-4. It is two comparable pulses which are delayed 2 units apart in CIR, and there is two deep fading points on index 16 and 48 in CFR. It is the simplest two-ray channel model, but it still crashes the original system, IEEE 802.11a.
8
8
-Figure 2-5 Two-ray channel model
Figure 2-5 show carrier error rate in two-ray channel model. We can observe although SNR approximate 60dB, there is two errors, highlight in figure, on spectrum,
Figure 2-6 Carrier error rate
- 9 -
9
-Chapter 3
The proposed algorithm
3.1 Pseudo Channel Block Diagram
The pseudo channel block diagram show in Figure 3-1.
Figure 3-1 Block Diagram
In Rx, we can class the tones into two categories, one is deep-fading tones, and another
10
10
-is non-deep-fading tones. After the “Sync.”(synchronize), if received signal -is preamble, then input the data to “Channel Estimator” for estimating CFR, and then feed CFR into
“Pseudo Channel Block”. In this paper, the channel model is fixed, and we assume the channel estimation is accuracy.
Else if the signal is payload, then we resolve the non-deep-fading tones at first by equalizer, ”QAM Decision Unit”, and feedback resolved data to “Pseudo Channel Block”.
The remainder, deep-fading tones, is resolved by “Pseudo Channel Block”.
3.2 Pseudo Channel Algorithm Process
In Figure 3-1 below, pseudo channel process have three steps. The first step is removed redundant AWGN. The second step is convolute pseudo channel. Final step is decision.
Before discuss the process, we have use a simple example for understanding the algorithm easily. In Figure 3-2, we transmit p t and follow ( )( ) d t in Tx. Because of we focus the deep-fading tones, so the index 16, 48 are discussed in the example. Therefore, we only descript the process by index 16. Figure 3-2 below, X16( )t is the 16th carrier in OFDM system and P ,16 D are the transmitted data. 16
Figure 3-2 A sample of transmitted signal and index 16 of signal
Figure 2-5 show the two-ray channel we focus in this paper. When the 16th carrier through pass the two-ray channel, it is effected that first path and second path delay 2 units
- 11 -
11
-and is added AWGN (Additive White Gaussian Noise). Figure 3-3 show effected signal by two-ray channel. In Rx, the received signal is shown in Figure 3-3 below, r t . Formal 16( ) expression is shown as:
[ ] [ ]* [ ] [ ] (3.1)
Observation, there is no signal power about transmitted data after 2 unit time and it is only AWGN. By the general OFDM system, we get a interval signal after GI (Guard Interval) which is only AWGN in the 16th carrier. Therefore, there is noise enhance on index 16 in two-ray channel.
3.2.1 Removed Redundant AWGN
In Figure 3-3 below, the redundant AWGN of the received signal is replaced by zeros, shown in Figure 3-4. But the previous signal should leave the deep-fading data by feedback signal, non-deep-fading data. And express as:
12
Figure 3-4 Removed redundant AWGN after 2 unit time
3.2.2 Convolute Pseudo Channel
Figure 3-5 Pseudo channel impulse response
At the first, we define pseudo channel for the two-ray channel. Figure 3-5 show the pseudo channel and express as:
39
If we convolute two-ray channel and pseudo channel, we can find the result which is delay apart one symbol length. Shown in Figure 3-6 and express as:
13
Figure 3-6 Convolute two-ray channel and pseudo channel
Now, convolute the receiver signal and pseudo channel, shown in Figure 3-7. And the expression is followed as:
0
Figure 3-7 Convolute the receiver signal and pseudo channel
14
14
-3.2.3 Decision
In Figure 3-7, the rectangle show reconstruction of signal which is deep-fading by two-ray channel. Therefore, we can decision the deep-fading signal after convolution pseudo channel. The first, we transfer y n by FFT and express as: ''[ ]
''[ ] [ ] [ ] 0[ ] (3.6)
Y k =S k −P k +W k
Then, estimate the data we resolve by previous data, P k . Express as: [ ]
'[ ] ''[ ] [ ] [ ] 0[ ] (3.7)
S k =Y k +P k =S k +W k
3.2.4 Noise Enhance
Observation, if we don’t execute the first step, removed redundant AWGN. The result is also Equation 3.7. But Figure 3-8 show the noise enhance on deep-fading point, 16 and 48.
Figure 3-8 Compare no removed and removed AWGN
15
15
-Chapter 4
Simulation results
In this chapter, to evaluate the pseudo channel algorithm, a typical OFDM system based on IEEE802.11a Wireless LANs, TGn Sync Proposal Technical Specification is used as the reference design platform. The performance of proposed algorithm is simulated under two-ray channel which we defined in section 2.2 with Additive White Gaussian Noise (AWGN).
In the simulations, the source data of transmitted length is 1024 Bytes and use the coding rate of 3/4 and modulation of 64-QAM. Follow as Table 4-1
Parameter Value System Platform IEEE 802.11a
Modulation 64-QAM
Coding rate 3/4
PSDU length 1024 Bytes
Carrier frequency 2.4 GHz
Bandwidth 20 MHz
Symbol rate 250 KHz
IFFT/FFT period 3.2 sμ
GI duration 0.8 sμ
Approx. packet duration 200 sμ
Table 4-1 Simulation environment
16
16
-In Figure 4-1, the no algorithm method is divergence, and only the pseudo channel algorithm is convergence on 51dB. Additional removed redundant AWGN is convergence on 36dB, it gain 15dB. Figure 4-2 shows the SNR vs. BER. Figure 4-3 shows the SNR vs.
PER.
Figure 4-2 BER vs. SNR
And the SNR vs. CER by pseudo channel algorithm with removed redundant AWGN on every carrier is shown in Figure 4-5. The influenced carrier decrease on about 30dB, and it convergent on 36dB. In Figure 4-6, CER which no algorithm is 0.98 on 36dB. And the pseudo channel with removed redundant AWGN of CER is 0.68. To observe, the expected value is good to 0.68 on deep fading carriers, and the FEC enable to decode.
17
17
-Figure 4-3 PER vs. SNR
Figure 4-4 SNR vs. CER by pseudo channel algorithm with removed redundant AWGN
18
18
-Figure 4-5 SNR vs. CER on 36dB by pseudo channel algorithm with removed redundant AWGN
19
19
-Chapter 5
Conclusion and Future Work
5.1.1 Conclusion
This thesis proposes a pseudo channel method. The deep fading carrier is fading by resonance phenomenon in two-ray channel. It is all zero value by sum of peak and trough.
And then, the ADC in receiver sample zeros on deep fading carriers.
Therefore, find the remaining energy of deep fading carriers is main ideal of the algorithm. And we find the GI have the remaining energy, but complex with previous data.
The feedback mechanism should be used for cancelling previous data.
Although, it works to reconstruct the deep fading carriers, but the noise enhance is a big problem. Because of total energy in GI is less than one FFT length. The SNR loss occurs.
The removed redundant AWGN is proposed for gain SNR. And the performance gain 15dB against no removing. But there is still 23dB gap.
A common sense is up sample in ADC, and it get more non-zero values in GI on deep fading carrier.
20
20
-5.1.2 Future Work
The next study is enhancing performance by MIMO-OFDM system. It is simple idea that the number of received antenna will gain the same number time of performance. But the preliminary studying find it is better than the idea. Because of the STBC is a diversity which changes the channel type. Therefore, the deep fading effect in two-ray channel is destroyed and the noise enhance don’t occur or less.
21
21
-Bibliography
[1] Xiang-Gen Xia, “A New Channel Independent Precoded OFDM Systems Robust to Spectral Null Channels”, 2000 IEEE.
[2] Jian Wang, Jian Song, Zhi-Xing Yang, Lin Yang, and Jun Wang, “Frames Theoretic Analysis of Zero-Padding OFDM Over Deep Fading Wireless Channels”, 2006 IEEE.
[3] John Terry and Juha Heiskala, “OFDM Wireless LANs: A Theoretical and Practical Gide”
[4] Simon R. Saunders, “Antennas and Propagation for Wireless Communication System”
[5] Kaveh Pahlavan and Allen H. Levesque, “Wireless Information Networks”
[6] Tzi-Dar Chiueh and Pei-Yun Tsai, “OFDM Baseband Receiver Design for Wireless Communications”
[7] Simon Haykin and Barry Van Veen, “Signals and Systems”
[8] Andrea Goldsmith, “Wireless communications”